Ideal Green Wall Locations: Melbourne

Authored by: Ryan Waites and Julian Cape

Duration: 120 mins 

Level: Beginner         Pre-requisite Skills:Python

Scenario

1. As a city planner, I want to determine the best potential locations for new green walls in the city of Melbourne. What locations would be most beneficial?

Green walls have been proven to have a moderating effect on multiple environmental factors. In addition to this they provide wildlife corridors and visual amenity. I would like to visualise a range of relevant factors such as pollution, temperatures and tree coverage in order to determine the most impactful locations for the installation of new green walls around Melbourne.

What this use case will teach you

At the end of this use case you will:

A brief introduction to the issue and datasets used

Issue being addressed

It is a priority of the City of Melbourne to create a healthier natural environment and adapt the city to climate change. These goals are outlined in the council's visions and goals.

One component of their plan to tackle these issues is Melbourne's Urban Forest Strategy. The strategy aims to improve the overall ecological health of the city by addressing canopy cover, plant diversity, vegetation health and ecology.

Green walls are vertically built structures intentionally covered by vegetation. They provide an extra option to assist in meeting several of these goals, enabling greening of areas with little available space. Several considerations, requirements and benefits associated with green walls can be explored further in the citys growing green guide. The datasets explored in this analysis provide insight into specific locations throughout the city where green walls are feasible and can have the greatest impact.

Datasets overview

All of the datasets used in this analysis come from the City of Melbourne's open data portal.

The first dataset to be analysed is that of the city's Microclimate sensors, which have been placed in several locations around the city. These provide a host of relevant environmental data including airborne particulate matter and ambient temperature. We will access this information (Microclimate Sensor Readings) and pair it with its location dataset (Microclimate Sensor Locations) to map the results and determine where green walls will have the greatest climatic effect.

We then perform the same process with the Pedestrian Counting System - Monthly (counts per hour) dataset and its pair Pedestrian Counting System - Sensor Locations to find the most heavily trafficked areas, where extra visual amenity will be seen by the most people.

The analyses conclude by plotting the citys tree canopy data and comparing it with butterfly and insect surveys. These show the direct relationship between canopy coverage and ecological diversity, which in turn highlights the need for green walls in areas with the least amount of canopy.

Package and data imports

To begin we will import the required libraries and datasets to perform our exploratory analysis and visualisation of the datasets.

The following are core packages required for this exercise:

If you attempt to run this first cell and there is a 'module not found' error, you may need to install the package on your system. Try running: pip install -package name- to install the missing package. If this doesn't work, you may need to try Google!

Particulate Matter Analysis

Information regarding standards for PM10 and 2.5:

https://www.epa.vic.gov.au/for-community/environmental-information/air-quality/pm10-particles-in-the-air

https://www.epa.vic.gov.au/for-community/environmental-information/air-quality/pm25-particles-in-the-air

Retrieve all data from PM10 24 hour average sensors, with erroneous values removed, from within the last year.

With datatypes sorted, view the total entries from each site and their mean PM10 recordings.

Comparing the data to the standards

With the annual average national standard at 25 µg/m3 (Victorian standard 20 µg/m3), we can see that 9 of 13 sites have exceeded it and all are exceeding the Victorian standard. However, theres a varying amount of data collected for each site so we should check the first and last recorded dates.

As evident above, it appears no sites actually have all of the last 12 months worth of data, making annual averages difficult to rely upon without looking further back for a better overall view.

Lets obtain all available data and look at the averages by months and see if we can see any clear trends that indicate a likely reduction of these averages within a yearly timeframe.

Interpreting the graph

We can see all sites generally follow the same trend however based on current and past data there's no reason to expect any significant drops that would reach Victorias annual standard. Of the sites still currently recording data, sites 1009, 1012, 1016 & 1014 are showing the highest averages.

Looking purely at the last year of data

Mapping sensor data

Let's now look at merging our data with the sensor location dataset so we can visualise the results.

Sensor location dataset: https://data.melbourne.vic.gov.au/Environment/Microclimate-Sensor-Locations/irqv-hjr4

Pedestrian data analysis

Relevant Dataset

Pedestrian Counting System - Monthly (counts per hour). Retrieve all data recorded since the 11th of April 2021 to help identify high pedestrian traffic areas.

Calculating hourly averages

With columns and datatypes sorted, lets focus on the average hourly pedestrian count for each sensor.

Pairing datasets

Lets now obtain the paired dataset containing each sensors location.

Observations

With our data merged, can see below that only one sensor (ID: 79) was installed during the period of data we're examining, we should take note of this as it could factor into its average. Otherwise we have a complete timeframe.

Insect Analysis

Mapping Insect Density

This analysis combines three MOP datasets to calculate the relationship between tree canopy and number of different insect species present at given locations. This relationship provides further justification of the green wall locations chosen, as the locations are chosen from areas around the city with low tree canopy coverage.

First we load and format the relevant datasets:

Creating shapefile to match monitoring locations

The insect and butterfly data only has the names of the monitoring locations listed for its geographical data. We will create a shapefile with coordinates taken from an online map. Including the correct coordinates reference system (CRS) is essential for the points to be in the right place.

Creating Final dataset

Now we will combine the three datasets to determine the amount of insects counted at each location in relation to the area of canopy cover found within 50, 100 and 200m of the monitoring locations.

Graphing the data

Now that we've combined our insect and tree canopy datasets, we can compare the results to view the relationship between number of insect species and tree canopy coverage.

Results

We can see from the graphs that as we expected, there is a strong relationship between tree canopy coverage and number of insect species present at a location. In our analysis, the relationship becomes stronger when a wider area is considered for canopy coverage. The correlation between count of insect species and canopy coverage ranged from 0.47 when considering canopy upto 50m from the monitoring site to 0.77 when considering upto 200m.

0.77 indicates a strong linear relationship.

0.47 still indicates a relationship, but the results are not as decisive.

Intuitively it makes sense to consider the wider area, as many flying insects would have a range of at least 200m. Some travel thousands of kilometres!

These results reinforce the need for green walls at locations with little canopy coverage.

Mapping the results

Tree Canopy Area

First we'll prepare the canopy data for the map, to highlight the areas with less vegetation. The canopy data currently covers an area much bigger than what we're interested in, which is the CBD. We'll create a polygon and trim the data to the CBD.

Pedestrian Traffic

Here we visualise the hourly average for pedestrian traffic via scaled blue circles. Large circles indicate busier locations.

PM10 Readings

The following code visualises the paired 24 hour averaged sensor data we obtained above. Circle colour (green to red) is representative of the sensors reading average.

NOTE:

You may need adjust your location in the street view tab by clicking the nearest street in the bottom left-hand map to avoid viewing user-generated google maps images.

Identified ideal green wall locations

The above analysis and visualisations allow us to identify ideal green wall locations around the city of Melbourne.

This analysis is by no means exhaustive, and there are many other aspects to consider. Once a potential area was identified using the map above, Google Street view was used to confirm an area's suitability.

The main driving factor in identifying locations was the absence of tree canopy. Nearby high levels of particulate matter confirmed a location as ideal. Areas with both high and low levels of pedestrian traffic could be seen as ideal, as:

  1. Green walls increase the visual amenity of an area, increasing the number of people using certain routes.
  2. Installing green walls in areas with already hgih levels of traffic ensure many people are able to enjoy them.

Some preliminary results are seen below:

Location 1

Corner of Flinders St & Elizabeth St, nearby to two of our highest PM levels (28.08, 28.43) and a well foot trafficked area (687) with area for large green wall. View

Location 2

Corner of Lonsdale St & Elizabeth St, near to another high PM level (28.83), though with varying degrees of foot traffic in its surrounding areas (171, 178, 511), it presents another good opportunity for a smaller green wall. View

Location 3

Drewery Ln off Lonsdale St offers a variety of sections for green walls & vertical gardens whilst being nearby to some high traffic areas (556, 411, 549) and within 2 blocks of a high PM level reading (28.83). View

Location 4

Sugden place off of Little Collins Street. There is very little vegetation in this part of the city and there is a large span of bare wall at this location. Installing a green wall here would provide wildlife with a valuable refuge. View

Location 5

Goldie Pl off Lonsdale St presents a large unused wall space that could serve as a large greenwall, located only streets away from a high PM level (28.83). View

Location 6

Another location on Little Collins Street. The unvegetated nature of the street would be greatly improved with multiple green walls, increasing visual amenity and attracting foot traffic. The sensor to the east has a mean PM10 reading of 27.5. View

Location 7

This location is on the corner of Little Collins and Collins Way. The walls and roof of this Woolworths Metro could support a wealth of plants, further revitalising the street. A pedestrian sensor around the corner records low levels of foot traffic. This could be improved with greater visual amenity in the area. View

Mapping the identified green wall locations

Now that we've identified seven ideal locations, let's plot them on a map and see what they look like in Google Street View.

We hope you've enjoyed this use case, and are inspired to use it to explore greening options for the city!